First-Principles Study of the Adsorption and ... - ACS Publications

Subscriber access provided by ST FRANCIS XAVIER UNIV

C: Surfaces, Interfaces, Porous Materials, and Catalysis

First-Principles Study of the Adsorption and Depolymerisation Mechanism of Sodium Silicate on Iron Surface at High Temperature Nam Van Tran, Anh Kiet Tieu, Hongtao Zhu, Huong Thi Thuy Ta, Thi Dinh Ta, and Ha Manh Le J. Phys. Chem. C, Just Accepted Manuscript • DOI: 10.1021/acs.jpcc.8b06073 • Publication Date (Web): 18 Aug 2018 Downloaded from http://pubs.acs.org on August 21, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 32 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

First-principles Study of the Adsorption and Depolymerisation Mechanism of Sodium Silicate on Iron Surface at High Temperature Nam V. Tran, A. Kiet Tieu, Hongtao Zhu*, Huong T. T. Ta, Thi D. Ta, Ha M. Le School of Mechanical, Materials, Mechatronic and Biomedical Engineering, University of Wollongong, Northfield Avenue, Wollongong, NSW 2522, Australia *

Corresponding author: [email protected]

Abstract Silicate glasses are potential materials for lubrication at elevated temperature due to their exceptional thermal stability, low cost and environmentally friendly properties. Although the frictional behaviours and characterizations of silicate glass tribofilms have been studied by a number of experiments, the formation mechanisms as well as the chemical insight into the iron – silicate tribofilm have still remained unclear. In the present study, the adsorption and depolymerisattion of sodium sorosilicate cluster Na6Si2O7 on Fe (110) surface have been studied using both first principles molecular dynamics (FPMD) and density functional theory (DFT). Comparisons of adsorption processes and electronic structure of some typical configurations at different temperatures have been carried out. The results strongly suggest that silicate cluster chemically adsorbs on the Fe (110) surface by forming multiple Fe-NBO bonds. The iron surface plays an important role in the dissociation of the non-bridging oxygen (NBO) by significantly reducing the strength of Si-NBO bonds. Electronic structure calculation reveals that the charge transfer between the lubricant and the iron surface during the adsorption may lead to a flow of alkali ions and a layering process in the tribofilm. The depolymerisation is observed at higher temperature and has larger activation energy compared to SiNBO dissociations indicating the process is more difficult. Temperature is also an important factor that contributes to the dissociation of NBOs as well as the depolymerisation of the lubricant, both of which can have several impacts on the lubricity of the tribofilm. This study provides a detailed understanding of the chemical reaction of sodium polysilicate on iron surface at high temperature.

ACS Paragon Plus Environment

1

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 32

1. INTRODUCTION The critical requirement of high temperature tribological systems is challenging the limitation of current lubricants. An improvement in the lubrication technology at elevated temperature conditions is necessary than ever. In recent years, silicate glasses have appeared as bright candidates for lubrication in harsh conditions due to their exceptional thermal stability, low cost and environmentally friendly properties 1. This kind of lubricant can provide superb friction reduction and anti-wear properties at high temperatures

2-7

. In addition, silicate-based compounds have self-restoration

capabilities which can repair the mechanical damage at the bearing surfaces 5. Although a number of experiments have been carried out to investigate the frictional behaviours and characterizations of silicate glass tribofilm at macroscopic scales, the formation mechanisms as well as the chemical insight into the iron – silicate tribofilm have still remained unclear. It has been found experimentally that alkali silicates of sodium or potassium can create a metallic silicate lubricating film by reacting with metallic rubbing surface 8. The film not only can protect the metallic surfaces but also can avoid local overheating thanks to the high specific and latent heat. The use of the silicate glass as a lubricant at high temperature was studied for the first time by M. Lambert and Sejournet 9. In the study, the glass powder has been used to provide lubrication between the die and the billet. At high temperature silicon in the silicate glass can react with iron oxide to create a eutectic mixture of Fe2SiO4

10

. The silicate compound then sticks on the oxide surface playing as a

protective layer. In general, melted silicate lubricants adsorb on the rubbing surface through chemical or physical absorption. The interaction between lubricant molecules and surfaces plays an important role in controlling the performance of the tribofilm. It has been proven that the anti-wear capability of the lubricant is influenced by the interaction strength of the lubricant−surface 11. Therefore, in order to optimize the tribofilm formation process, the understanding of the adsorption and reaction between silicate lubricant and metal surfaces at an atomic level is required. Despite the growing interest in understanding the atomistic mechanisms for the low friction and anti-wear of silicate glasses, our present knowledge of this novel lubricant at high temperature is still limited. Computational simulations have been used successfully to investigate the interactions as well as tribo-chemical reactions between surface and lubricant for a long time. One of the most common method used to investigate tribological system is Molecular Dynamics (MD)

12

. Due to the large

number of atoms involved in the tribological systems, MD simulations can take into account the effects of large-scale features such as steric effects due to the roughness of rubbing surfaces Ta et al.

13

13

. In particular,

studied the influences of loading pressure and copolymer concentration on the structural


2

Page 3 of 32 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


properties and tribological performance of aqueous copolymer lubricant in mixed lubrication regime. Zhou et al. 11, 14 used MD simulation to prove the relationship between the cohesive energy in the selfassembled monolayers (SAM) and anti-wear performance. Although MD simulation is very powerful to describe the dynamics of tribological systems, the use of classical force fields in MD is unable to describe properly tribochemical reaction. This is due to the fact that the method cannot describe the electronic properties such as electron transfer, Pauli repulsion, mutual polarization, electron exchange and correlation. In addition, since the description of the local interactions in MD is predetermined by the parametrization of the applied force fields, the method requires the parameterization for every new system 15. On the other hand, first principle simulations with density functional theory (DFT) are more useful tool when dealing with tribochemical reaction since they can describe accurately the electronic properties of a system. There are a number of publications that have successfully used DFT simulations to investigate chemical reactions in tribological systems. For example, Migala et al. reported the interaction of the basic structural element of silicates (silicate group SiO4) and iron surface 16-17. Jaiswal et al. proposed the relationship between the HOMO (Energy of Highest Occupied Molecular Orbital) and LUMO (Energy of Lowest Unoccupied Molecular Orbital) levels and the anti-wear performance of several lubricant additives on iron surface, and their results were in good agreement with the anti-wear performance obtained from experiments

18-19

. Furthermore, reaction pathways, transition states and

possible products of several lubricants adsorbed on iron surface have been studied by Righi and colleagues, which provided a deep insight into the tribochemical reaction processes 20-21. However, one major drawback of static DFT method is that it cannot simulate the dynamics of lubricants at high temperature. In addition, due to the high computational resource the sizes of the tribological systems are often limited. First-Principles Molecular Dynamic (FPMD) is another powerful tool to investigate tribochemical reaction at elevated temperature. Deriving from first-principles, the method can accurately describe the time and temperature evolution of the chemical events and the dynamic behaviours. Due to these advantages, the number of publications using FPMD to investigate chemical systems is increasing in recent years

22

. For instance, Mosey et al. introduced several thermal

decomposition pathways that involves in the loss of radicals, olefins, and sulfides which led to the components

of

anti-wear

metathiophosphates (MTPs)

23

film

precursors

of

zinc

dialkyldithiophosphates

(ZnDDPs),

. In another publication Mosey reported a cross-linking process of zinc

phosphate (ZPs) under compression

24

. Furthermore, Righi et al. applied FPMD to study the

fundamental mechanism for carbon-film lubrication by water

25


3

. Thus, a combination of DFT and


Page 4 of 32

FPMD can provide new useful information to understand the mechanism of tribochemical reactions of new lubricants in terms of both dynamics and electronic structure. In practice, the oxide layer on the iron surface can be abraded during shearing process leaving the chance for the lubricant to react directly with the nascent iron surface 26-27. In addition, Fe from the surface can diffuse into the glassy layer through cracks or gain boundaries and reacts with the silicate therein

28-29

. Therefore, in this study we use both DFT and FPMD to investigate the behaviour of the

sodium silicate cluster interacting with nascent iron surface at high temperature. The study will provide deep insight into the anonymous adsorption mechanism and anti-wear performance of silicate lubricant on iron surface. Our FPMD simulations will provide real-time monitoring of the adsorption processes. Meanwhile, DFT calculations will yield an in-depth understanding of the electronic structure of sodium silicate cluster and its adsorption on iron surface. Their electronic structure is also analysed to investigate the effects of chemical reactions on anti-wear reduction as well as the fundamental mechanism for the tribofilm formation.

2. COMPUTATIONAL DETAILS In order to investigate the interactions of sodium silicate lubricant with iron surface, we performed First-Principles MD (FPMD) simulations based on the Born-Oppenheimer method by using Vienna Ab Initio Simulation Package (VASP)

30-31

. The electron exchange and correlation effects are

calculated using the generalized gradient approximation (GGA) of Perdew−Burke−Ernzerhof (PBE) exchange and correlation functional

32-33

. The projector augmented-wave (PAW) method is used to

represent the interactions between valence electrons and ionic cores 34. The convergence criteria of 10-4 eV for electronic self-consistent loop and 10-3 eV for ionic relaxation were applied for all calculations. In our simulation procedure, there are three primary steps: i. Optimizing the structures of Fe (110) surface and sodium silicate cluster. ii. Performing FPMD simulations for the Na6Si2O7−Fe system. iii. Choosing several interesting adsorption configurations from the FPMD trajectories to produce DFT calculations and investigate the electronic structures. The following will detail how our simulations are set up. Optimizing the structures of Fe (110) surface and sodium silicate cluster. Iron surface was created by cleaving the unit cell of bulk Fe. In this research Fe (110) surface was chosen because they are the most thermodynamically stable surface

35

. A vacuum of 20 Å was

added along the z direction of the surface to prevent the interactions between the images of the simulation cell. The surface was modelled by a periodic supercell of 17 × 15 × 25 Å, containing three ACS Paragon Plus Environment

4

Page 5 of 32 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


atomic layers (Figure 1). The selection of the slab thickness was based on the work by Righi et al. 20, which indicated that a slab thickness of two atomic layers is sufficient to obtain the surface energy of iron. Sodium silicate was used as a lubricant to study the chemical reactions. It is noted that the polymer of sodium silicate usually exists as long chain, ring or complex networks. However, due to the high computational cost in static DFT and FPMD calculations, a small cluster (Na6Si2O7) was chosen as a representative of sodium sorosilicate (Figure S1 in Supporting Information). The cluster contains two groups of SiO4 and has both non-bridging and bridging oxygen atoms. Therefore, it is able to represent all types of bonds existing in the long-chain polymer. Although the small size of the current model limits its ability to explain the tribofilm formation of a thick glassy layer on the iron surface, the model can be used to predict the tendency of bond formation or dissociation at the interface between silicate glass and iron. Such information is critical to understand the adherence mechanism of silicate lubricant on iron surface in real systems. The strategy of cutting a long chain polymer into small molecules to study its interaction with well-characterized surfaces has been described by L. Delle Site et al

36

. The optimized structures as well as geometrical parameters of the sodium silicate cluster are

provided in the Supporting Information. Performing FPMD simulations for the Na6Si2O7−Fe system. After completing structure optimization, Na6Si2O7 cluster was placed at 5 Å above the Fe (110) surface as shown in Figure 1. This distance is relatively equal to the length of the cluster, making sure the molecule can freely rotate and approach the preferable site on the surface. During the FPMD simulations, the bottom iron layer was frozen, while the top two layers and the adsorbate were allowed to relax. The simulations were performed at six different temperatures (1000, 1100, 1200, 1300, 1400, and 1500 K) using the Nose-Hoover thermostat to maintain the temperature of the system. The total duration of each simulation was 10 ps corresponding to 20000 MD steps (with time step of 0.5 fs). It is worth mentioning that the selection of temperature range was based on experiments 37. Due to the large size of the supercell Brillouin zone integrations were sampled over the Γ point, and the plane-wave expansion of the electronic wave function is truncated by setting the cutoff energy to 350 eV. Due to the enormous computational effort associated with these FPMD simulations, only one initial configuration is considered in the study.


5


Page 6 of 32

Figure 1. Initial structure of Fe-Na6Si2O6 system. Blue (Fe), Red (O), Purple (Na), and Orange (Si). Performing DFT Calculations for the Selected Na6Si2O7−Fe Configurations In order to extract the electronic structures of the system which are useful to understand the interaction between iron surface and silicate lubricant. Standard DFT calculations have been performed on selected interaction configurations from the FPMD trajectories. After finishing the FPMD simulations, several selected interaction configurations are chosen to capture the significant changes in the adsorbed state of Na6Si2O7 cluster on the Fe surface. The self-consistent calculations were then carried out to analyse the electronic structure of the systems such as partial density of state (PDOS), charge transfer and bond overlap population (BOP). For higher accuracy, the cutoff energy and k-point mesh were increased to 400 eV and 3 × 5 × 1, respectively. In addition, due to the tendency to underestimate the band gap of GGA exchange and correlation when dealing with transition metals like iron

38

, A GGA+U with Ueff = 4 eV (U = 5 eV and J = 1 eV) was used to provides more accurate

electronic descriptions 40-41

39

. We used the density derived electrostatic and chemical (DDEC6) approach

to calculate atomic charges. Further details on the method and comparison with the Bader

approach can be found elsewhere 42.

3. RESULTS 3.1.

Electronic Structure of the Isolated Systems

In order to understand the interaction between silicate lubricant and iron surface, electronic structures including DOS and BOP have been calculated on the isolated silicate cluster and the iron ACS Paragon Plus Environment

6

Page 7 of 32 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


surface to create a reference. Figure 2 reveals the PDOS of the isolated silicate cluster (a) and the pure iron surface (b). At the isolated state, due to the localization of molecular orbitals, the PDOS of Na6Si2O7 appears as very sharp peaks, which are typical for a molecule or cluster model. The PDOS of Na6Si2O7 is ranging from -7.5 eV to -0.4 eV and mainly composed by Si 3s, Si 3p, and O 2p orbitals. In addition, there is a strong overlap between 3s and 3p orbitals of Si atoms and 2p orbitals from O atoms indicating strong bonds between them. The great contribution of Si 3s orbital is due to the fact that Si atoms are sp3 hybridized and have a tetrahedral configuration 43. The four sp3 orbitals have one electron each, which overlaps with 2p orbital of oxygen to form the covalent bond. The result is consistent with the calculated bond angle indicating O-Si-O angle is 109.49o (Table S1). It is noted that the O 2p peaks located right below the Fermi level do not overlap with either silicon or sodium’s states. Hence, those peaks must belong to the lone pair orbitals of oxygen. Due to the ability to donate electrons, those lonepair orbitals play an important role in the HOMO and LUMO interaction between the lubricant and the iron surface 18.

Figure 2. PDOS of the isolated silicate and the Fe surfaces: (a) isolated Na6Si2O7, and (b) the Fe (110) surface. Figure 2b shows the total and partial density of states for the nascent Fe (110) surface. It is well known that the d electrons are the major contribution to the total density of states (TDOS) of transition metals. In this case, we also observe that TDOS of Fe (110) surface are mostly composed of Fe 3d orbitals, the contributions of Fe 4s and 3p orbitals are negligible compared to the 3d orbitals. The Fe 3d state is distributed in the energy range from -7 to 4.5 eV. By comparison with Jun Hu's report

44

, we

found that by adding the on-site Coulomb interaction the PDOS of Fe surface has been expanded to above and below Fermi level by approximately 1 eV. In addition, the main peak below the Fermi level


7


Page 8 of 32

is split into smaller peaks. Apparently, Fe surface has a high density of 3d states that cross the Fermi level, so the d electrons participate actively in chemical reactions with the adsorbate. The covalent characteristic of bonds is examined by analysing the bond overlap population (BOP). The BOP was calculated by the integration of crystal orbital overlap population (COOP) analysis utilizing Local-Orbital Basis Suite Towards Electronic-Structure Reconstruction (LOBSTER) code 45-47 after the self-consistent calculation by VASP code. The method has been successfully used to evaluate chemical bonding from plane-wave based DFT calculations value means the higher level of covalent characteristic in a bond

49-50

48-49

. Basically, the larger BOP

. Based on our BOP calculation

(Table 1) the bonds between silicon and non-bridging oxygen (NBO/O) Si-O are covalent bonds with the averaged BOP value of 0.31e. Meanwhile the averaged BOP value for bonds between silicon and bridging oxygen (BO/Ob) Si-Ob is 0.24e. This result indicates that Si-NBO has a higher level of covalency than that of Si-BO. The BOP calculation is consistent with the measured bond lengths from both our calculation and experimental studies which reported that Si-Ob bond length (1.69 Å) is longer than that of Si-O (1.65 Å) (Table S1). Since the longer bond lengths indicate in general a more ionic bonding character in the Si-Ob bonds, one expects the BO atoms to have a lower BOP than the NBO atoms. Meanwhile, the BOP of O-Na and Fe-Fe bonds are very small indicating that those bonds do not show the covalent characteristic.

Table 1. Average Bond Overlap Population (BOP) of ionic and covalent bonds in the isolated Na6Si2O7 cluster and the Fe surface.

3.2.

Bonds

Si-O

Si-Ob

O-Na

Fe-Fe

BOP (e)

0.31

0.24

0.06

0.01

Interactions of lubricant on the Fe surface at high temperature

3.2.1. Adsorption of Na6Si2O7 on iron surface This section will discuss the behaviours of silicate cluster during the adsorption on iron surface at 1000 K which is the lowest temperature in the simulation range. Figure 3 reveals the snapshots of sodium silicate cluster on Fe surface during the simulation at 1000 K, the free energy profile during the simulation is provided in the Supporting Information. In order to have a better illustration, only the iron atoms on the top layer have been shown. The BO atom has been marked as Ob, while NBO atoms have been noted from O1 to O6.


8

Page 9 of 32 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 3. Different adsorption configurations of Na6Si2O7 on the Fe surface during the simulation at 1000 K. The snapshots were taken at 0.6 ps (a), 1.3 ps (b), 1.4 ps (c), and 2.5 ps (d).

Figure 3 shows that Na6Si2O7 molecule adsorbs on the Fe (110) surface from an early stage of the simulation. As the molecule gradually approaches the ion surface to find the favourable adsorption position, it rotates and forms the first bond between O1 and Fe at 0.6 ps (Figure 3a). The dangling part of the molecule shortly goes down and bonds to the iron surface at 1.3 ps (Figure 3b). In this configuration, the BO (Ob) also forms a bond with Fe atom. However, the Fe-Ob bond is weak, and it is dissociated quickly leaving only two NBOs bond with the surface (Figure 3c). At 2.5 ps the Na6Si2O7 molecule anchors to the Fe surface by three NBOs as shown in Figure 3d. The first SiO4 unit links to the surface using the bonds of two NBOs while the second one only uses one NBO for the bonding. In general, the trajectory analysis shows a tendency that the sodium silicate molecule prefers to form multiple Si-O-Fe linkages with the iron surface. The number of NBOs involved in the bonding with the iron surface gradually increases from one to three during the simulation. Table 2. Bond Overlap Population (BOP) of Na6Si2O7 cluster adsorbed on Fe surface at 1000 K*. Na6Si2O7 cluster

Fe-O

Time (ps)

Si1-O1

Si1-O2

Si1-O3

Si1-Ob

Si2-Ob

Si2-O4

Si2-O5

Si2-O6

Total

0.6

0.18(1)

0.33

0.36

0.29

0.20

0.29

0.32

0.32

0.48

1.3

0.15(3)

0.34

0.41

0.21

0.21

0.19(2)

0.33

0.37

1.06

1.4

0.16

(3)

(2)

0.37

0.34

0.97

2.5

0.25(2)

0.35

0.40

1.31

0.35

0.36

0.23

0.23

0.17

0.20(2)

0.42

0.27

0.24

0.15(3)

*The superscript indicates the number Fe bonded to NBO in Si-O. Table 2 presents the BOP values of four selected configurations above, the BOP of O1-Fe bond at 0.6 ps is around 0.48 e. This result demonstrates that O-Fe bonds have a strong covalent


9


Page 10 of 32

characteristic compared to the Si-O bonds of the isolated cluster (0.31 e). Furthermore, the Fe-O bond length is fluctuated between 1.59 and 2.39 Å with an average value of 2 Å (Figure S4a) which is close to the Fe-O covalent bond length (1.98 Å). The covalent bond length between Fe and O is estimated by the summation of the covalent radii of Fe (1.32 Å) and O (0.66 Å ) 51. Additionally, sodium atoms do not form any specific bond during the simulation. Apparently, the interaction of alkali ions is mainly through a weak polarization contribution

52

. Therefore, the Na ions can only form weak and non-

directional ionic bonds with NBOs. This weak ionic bond leads to the high mobility of sodium ions as shown in the mean squared displacement (MSD) in Figure S4b. The formation of O-Fe covalent bonds also indicates that the Fe – sodium silicate interaction is a result of chemical reaction rather than physical absorption. In addition, (BOP) calculation in the Table 2 shows that the covalent bond Si-O can be weakened due to the formation of Si-O-Fe linkage. At 0.6 ps when the O1-Fe bond was formed, the BOP of Si1-O1 bond decreased from 0.31 e to 0.18 e (a 42% drop). The calculated bond distances in Figure S3 also indicate that Si1-O1, Si1-O2 and Si2-O4 bond lengths fluctuate in a larger range compared to other non-bridging bonds after the bond formation with iron surface. This phenomenon can be explained by the fact that the O donates its electrons to form covalent bonds with the Fe atoms. Therefore, there are fewer shared electrons in the bond between Si and O that weakens the Si-O bond. As a consequence, it is easier to break the Si-O bond after the absorption. Moreover, the BOP calculation also shows that the reduction of Si-O bond strength is more profound when O absorbs on the threefold position of the surface. In particular, the BOP of Si1-O1 at 1.3 ps and Si2-O4 at 2.4 ps (O adsorbs on the threefold position) significantly drop to 0.15 e which is the smallest value in the four selected configurations. In addition, as O shares its electrons to form multiple O-Fe bonds, the average BOP of O-Fe bonds decreases as the number of OFe bonds increases (Table 2). However, the total BOP of O-Fe still increases over time, which proves the adhesion capability of the sodium silicate to iron surface.


10

Page 11 of 32 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 4. PDOS of the selected adsorption configurations at 1000 K. The selected configurations correspond to the snapshots taken at 0.6 ps (a) and 2.5 ps (b) in Figure 3. PDOS for the intermediate configurations at 1.3 ps and 1.4 ps can be found in the Supporting Information (Figure S5).

In order to provide more details in the iron-silicate interactions at electronic level, we calculated the PDOS for each selected snapshot configuration and compare them with the PDOS of the isolated systems. The PDOS for the configurations at 0.6 ps and 2.5 ps are shown in Figure 4 (The PDOS for the configurations at 1.3 ps and 1.4 ps can be found in Figure S5). Compared to the PDOS of the isolated system (Figure 2a), one can easily see that the PDOS of O and Si atoms in the adsorbed system become smeared and widened bands from -8.5 eV to 0 eV. Those broadened bands overlap with Fe 3d indicating that the Fe-O bond is a hybridisation between O 2p and Fe 3d orbitals. The PDOS of O becomes flatter as the number of O atoms bonded to the Fe surface increases (Figure 4b). In addition, the O 2p peaks are significantly shrunk, especially those peaks near the Fermi level, which suggests electrons are withdrawn from oxygen atoms. Furthermore, as the number of adsorbed oxygen increases, the O 2p states are shifted to lower energy levels. Figure 4b shows that the O 2p states mainly located between -7 eV to -4 eV. This result indicates a stronger bonding interaction caused by the strong overlap between O 2p and Fe 3d states. Apparently, the calculated PDOS is consistent with the BOP calculation showing O atom can form a strong covalent bond with Fe atoms on the surface.


11


Page 12 of 32

Figure 5. Charge transfer for the selected adsorption configurations at 1000 K. ρFe, ρO, ρNa, and ρSi are the total charge transfer of the Fe, O, Na, and Si, respectively. In the horizontal axis, (a-d) correspond to the selected configurations in Figure 3.

DDEC charge analysis (Figure 5) shows a consistency with the calculated PDOS, indicating a charge transfer from O to Fe atoms on the surface. The amount of electron transfer is calculated by comparing the net charges of the selected configurations with those of the isolated systems (the calculated net charges are provided in the SI). The positive values in Figure 5 indicate electrons are drained from the atom and vice versa. Obviously, the top layer is the most affected layer by the interaction. The data in Figure 5 shows that the Fe surface receives 0.441 e when the first Fe-O bond is formed, in which the top layer gains 0.319 e while the second layer receives 0.187 e (Table S3). Meanwhile, there is almost no change for the net charge of the bottom layer of iron surface. At the same time, O atoms lose 0.852 e overall, with the additional electrons being transferred to Si and Na atoms (0.337 e in total). As the more NBOs anchor to the surface, the more electrons are drained from O atoms while the total net charge of the surface becomes more negative. At 2.5 ps, the Fe surface receives 1.073 e in total (Figure 5), with 0.887 e of which being transferred to the top layer (Table S3). Meanwhile, O atoms lose 2.246 e and ~52% of these electrons are transfer to Si and Na. 3.2.2. Dissociation of Na6Si2O7 on iron surface This section will discuss about the dissociation process of Na6Si2O7 molecule on the Fe surface. The details of structures observed during the simulations at temperatures above 1000 K are discussed to provide the general description of the reaction paths. The electronic structures of several snapshots will be analysed to clarify the effects of Fe surface on the adsorption as well as the dissociation of Na6Si2O7 cluster. ACS Paragon Plus Environment

12

Page 13 of 32 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 6. Different interaction configurations of Na6Si2O7 on the Fe surface. The top panel (a-d) shows the snapshots taken at 1100 K. The bottom panel (e-h) shows the snapshots taken at 1400 K. Figure 6a-d illustrates the structures of the system observed during the simulation at 1100 K. Similar to the simulation at 1000 K the Na6Si2O7 cluster quickly goes down to create a O1-Fe bond between NBO/O1 and Fe at 0.4 ps. It is interesting to note that the O5 forms a covalent bond with Si1 atom creating the second BO (Ob2) at 0.5 ps. Importantly, the bond between Si1-O1 is dissociated immediately after the formation of Si1-O5 at 0.7 ps (Figure 6c and Figure 7a) suggesting that the formation of the Si1-O5/Ob2 bridging bond has a strong sequel on the bond strength between Si1-O1. In fact, the dissociation of the Si1-O1 bond immediately after the creation of the Si1-O5 bond helps maintain the 4-fold coordination of Si1. This result is consistent with experiment observed by Wu et al. 53

who found that Si in silica glass has 5-fold coordination only at extremely high pressure (>12 Pa).

Furthermore, the Si1-O1 bond breaking is consistent with the previous BOP calculation in Table 2 demonstrating that the adsorption of sodium silicate molecule reduced the strength of Si-NBO bond at which O is anchored with Fe. The bond between Ob and Si2 atoms is weakened with the BOP value of 0.15 e at 0.7 ps (Table S4), therefore the bond is quickly dissociated at 0.9 ps as showed in Figure 7a. It is noted that the dissociation of Si2-Ob broken the tetrahedral structure of Si. Therefore, Si can create a bond with Fe atom at 3.1 ps (Figure 6d). During the rest of the simulation we observed the recombinations of the Si2-Ob bond at 5.9 ps (Figure 7a).


13


Page 14 of 32

Unlike the trajectories at 1100 K we do not observe any formation of the second BO/Ob2 from the simulations at 1200 K and 1300 K. This could be explained by the fact that the system is exposed to higher temperatures, the cluster would follow different trajectories that does not necessary to form the second non-bridging bond. However, it is also worth mentioning that only one initial configuration is consider in the FPMD simulations due to the high computational cost. Therefore, it is still possible to observe the formation of the second non-bridging bond if the sample of initial configurations and the simulation time are sufficient

54

. Nevertheless, the procedure would require different method with

lower computational cost, i.e., reactive molecular dynamic 54 which is beyond the scope of this report. As mentioned above, the dissociation of the Si-NBO at 1100 K is caused by both surface effect and the formation of the additional bridging bond (Figure 6c). The absence of the second BO/Ob2 in the simulations at 1200 K and 1300 K makes it difficult for the Si-NBO bond to break. Therefore, the result at 1200 K and 1300 K also demonstrated that the additional bridging bond has strong impact on the dissociation of the Si-NBO bonds. Due to the similarity with the simulation at 1000 K, the snapshots for the trajectories at 1200 K and 1300 K are presented in the Supporting Information (Figure S7 and Figure S8). Figure 6e-h shows the structures of the system observed during the simulation at 1400 K. Due to the strong effect of temperature (at 1400 K), the Si-O bond in the Si-O-Fe linkage can be broken without the formation of the second BO/Ob2. The Si-O bond breaking took place at around 2.9 ps as shown in Figure 6g and Figure 7b. After the bond breaking, the dangling part of the cluster goes down to create a bond between Si1 and Fe on the surface as shown in Figure 6h. It is worth mentioning that the dissociated O at both 1100 K and 1400 K tends to adsorbs on the three-fold position. Therefore, the result suggests that three-fold position is the most stable absorption site of the dissociated O on the Fe (110) surface.


14

Page 15 of 32 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 7. Evolution of Si-O bond length in Na6Si2O7 cluster during the simulation at 1100 K (a) and 1400K (b).

Figure 8. PDOS for the selected adsorption configurations in Figure 6. The top and bottom panel show PDOS at 1100 K and 1400 K, respectively. Due to the similarity, only PDOS of the last three snapshots on the top (Figure 6b-d) and bottom (Figure 6f-h) panel of Figure 6 are shown. PDOS for the first two snapshots can be found in the Supporting Information. In addition to the geometry information, the electronic structure also reveals how Na6Si2O7 adsorbs and dissociates on the surface. Figure 8 shows the transformation of PDOS during the ACS Paragon Plus Environment

15


Page 16 of 32

simulation. The formation of the Si1-O5 and Ob-Fe (Figure 6b) at 1100 K does not result in a significant change in the system PDOS as these bonds are weak (Figure 8a and Figure S6). When the Si1-O1 bond is dissociated, however, the O 2p states below the Fermi level increase dramatically (Figure 8b). This is because the molecule in this configuration only forms a weak bond between Ob and Fe. Therefore, the bonding interaction is reduced making low energy states shifted up and merged to higher energy states. Due to the significant reduction of the bonding interaction, the peaks below -8 eV disappear (Figure 8b). Similarly, the degeneration of the overlap states below -8 eV has been observed in the case of 1400 K when Si1-O1 bond is dissociated (Figure 8e). However, the PDOS still shows an overlap peak below -8 eV, which is due to the stronger bonding interaction of O2 atom with Fe surface than that of Ob. The tendency of forming multiple bonds between O-Fe has been also found the two simulations at 1100 K and 1400 K, which results a similar trend for the O 2p states as mentioned in the section 3.2.1. The calculated PDOS (Figure 8 and Figure S9) shows that the intensity and the position of O 2p peaks are reduced and shifted to lower energies indicating strong bonding interaction and electrons are drained from the atom. The outcome is consistent with the DDEC charge analysis (Figure 9) which indicates that O atoms transfer electrons to the surface during the adsorption. Particularly, at 1100 K the O atoms lose a total of 2.722 e in comparison with the isolated system. Meanwhile, a similar amount of 2.701 e has been drained from O atoms in the case of 1400 K. The formation of the Si-Fe bond also modifies the PDOS of the system. Generally, the Si 3s and 3p orbital are smeared and shifted to below -4 eV (Figure 8c and 8f). The result indicates a strong interaction between Si and Fe atoms. The calculated BOP (Table S4 and S6) indicates Si forms a strong covalent bond with Fe atoms, the BOP of those bonds vary from 0.30 – 0.46 e. The DDEC charge analysis shows that Si receives the largest amount of charge when the Si-Fe bond is formed, which implies that Si atom withdraws electrons from the surface (Figure 9). Particularly, a total amount of 0.830 e and 0.931 e has been transferred to Si atoms in the simulation at 1100 K and 1400 K, respectively. The effect of Si-Fe bonds to the surface net charge is more significant at 1400 K compared to that of 1100 K, i.e., the formation of those Si-Fe bonds causes a loss of 0.247 e from the surface as shown in Figure 9b. Meanwhile, the loss of electrons from the surface has been compensated by a large amount of charge from O atoms at 1100 K. In general, the dissociation of Si-NBO bond leads to the formation of the Si-Fe bond, which in turn reduces the amount of charge transfer to the iron surface.


16

Page 17 of 32 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 9. Charge transfer for the selected adsorption configurations at 1100 K. ρFe, ρO, ρNa and ρSi are the total charge transfer of the Fe, O, Na and Si, respectively. In the horizontal axis, (a-h) correspond to the selected configurations in Figure 6.

3.2.3. Depolymerisation of Na6Si2O7 on Iron Surface As mentioned in the section 3.2.2 the Na6Si2O6 cluster only dissociates its NBOs but the backbone of the molecule does not change. Since the cluster directly bonds to the surface through the NBOs, those Si-NBO bonds are mostly affected by the surface. In addition, the formation of the additional Si-Ob2 bridging bond is also an important factor that causes the dissociation of the NBO. In this section, when the temperature reaches 1500 K, the influence of temperature on the system is more significant and we can observe the depolymerisation process in this condition as shown in Figure 10.


17


Page 18 of 32

Figure 10. Different snapshots of Na6Si2O7 on the Fe surface during the simulation at 1500 K. The snapshots were taken at 0.5 ps (a), 2.2 ps (b), 2.7 ps (c), 3.2 ps (d), and 3.4 ps (e). The snapshots in Figure 10 show a similarity with the previous simulations as the molecule tends to rotate and adsorb on the iron surface by one of its SiO4 group at the beginning of simulation. However, the dangling part does not quickly go down to the surface as we observe in the previous simulations. At 2.2 ps, the Si2-Ob is broken separating the molecule into individual SiO3 and SiO4 groups (Figure 10b and Figure 11). SiO4 group remained adsorbed on the surface while SiO3 group quickly goes down and anchors to the surface at 3.2 ps (Figure 10d). In the following steps, NBOs in Si1-Ob and Si2-O5 bonds are quickly dissociated from those SiO3/SiO4 groups leading to the bond formation between Si and Fe at 3.4 ps (Figure 10e). During the rest of the simulation the Si1-O2 bond is dissociated at around 7.5 ps (Figure 11). In general, the simulation at 1500 K indicates that the lubricant can be decomposed into small fragments, suggesting the existence of the depolymerisation process in harsh condition. In addition, both bond dissociation and depolymerisation occur, which is totally different from the lower temperatures when only bond dissociation observed. This result suggests a vital contribution of temperature in the shortening of the silicate chains. However, it is worth ACS Paragon Plus Environment

18

Page 19 of 32 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


to mention that the current simulation did not to account the effects of shearing rate and severe loads which are believed to cause the polymerization process at high temperature as observed in the case of ZDDP 24.

Figure 11. The evolution of Si-O bond length in Na6Si2O7 cluster during the simulation at 1500 K. The calculated PDOS in Figure 12a and S10 shows an increase in the Si 3s and 3p state at 6 eV and 4 eV after the Si2-Ob bond breaking at 2.2 ps. The result might be caused by two reasons. First, the depolymerisation transforms the BO into NBO. Since the bonding interaction between Si-NBO is stronger than that of the Si-BO, the higher energy states are down-shifted and merged to the lower states causing the increase in the intensity of the Si states. In addition, the SiO3 fragment behaves like an isolated cluster making its states more localized and become sharp peaks. Secondly, Si atoms can drain electrons from O after the bond dissociation. In fact, the Si2 atom in the SiO3 fragment gives less electrons to it surrounded O atoms. Our calculated charges indicate that the Si atoms receive 0.298 e after the depolymerisation. After the adsorption of the SiO3 fragment, the PDOS of Si and O states become more broadened (Figure 12b). As the adsorption of the SiO3 fragment increases the number of Si-O-Fe linkages, an additional hybridized peak below -8 eV is observed in the PDOS at 3.2 ps. At 3.4 ps, the Si1-Ob and Si2-O5 bonds are dissociated causing the disappearance of the additional hybridized peak. Moreover, the Si 3s states are strongly shifted to the lower energy level due the covalent bonds with Fe atoms on the surface. As show in Figure 12c, the Si 3s states are mainly redistributed at the energy below -8 eV. In addition, the Si 3p states become more smeared and widened bands. The result is consistent with the calculated BOP in Table S8 which indicates that Si forms strong covalent bond with Fe surface (the total BOP of Si2 atom with Fe surface is 0.53 e). A strong bond between Si and Fe was also reported by Bui et al. 55 when study the adsorption of Fe on silicene monolayer. According to their study, the 3d ACS Paragon Plus Environment

19


Page 20 of 32

shells of Fe and 3p shells of Si overlap significantly, which consequently results in a strong bonding interaction. PDOS and DDEC analysis (Figure 13) also reveal a charge transfer from O to the surface although more intensive compared to the previous simulations. The more significant charge transfer is due to the increasing number of O and Si bond to the surface. Apparently, O loses 3.488 e (Figure 13) making its total net charge increase to -4.959 e (Table S9) while Si atoms also witness a large charge accumulation (1.391 e). It is worth mentioning that the amount of charge received by iron surface is not higher than the previous simulations. This is due to the fact that Si atoms also drain electrons from iron surface when Si-Fe bond is formed. However, Fe surface still gains charges during the whole process.

Figure 12. PDOS for the selected adsorption configurations at 1500 K. The selected configurations correspond to the snapshots taken at 2.7 ps (a), 3.2 ps (b), and 3.4 ps (c) in Figure 10. PDOS for the configuration at 0.5 ps and 2.2 ps can be found in the SI (Figure S10).

Figure 13. Charge transfer for the selected adsorption configurations at 1500 K (unit e). ρFe, ρO, ρNa and ρSi are the total charge transfer of the Fe, O, Na and Si, respectively. In the horizontal axis, (a-e) correspond to the selected configurations in Figure 10.


20

Page 21 of 32 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


It is noticed that in all of simulations the dissociated NBO atoms tend to adsorb on the threefold position on the iron surface. The NBO atoms stopped moving from its position after the dissociation indicating that this position is very stable. The interaction and adsorption of oxygen on Fe (110) surface have been studied by both experiment and theoretical calculation for a long time. Currently, there is a controversial report about the most stable adsorption site of O atom on Fe (110) surface. Previous studies suggested that the long bridge sites are the most favourable position for O on Fe (110) substrate 56-57

. On the other hand, some recent studies reported the higher stability of superstructures with O in

the threefold coordinated hollow site 58-60. Our result supports the second finding as O atom prefers to adsorb on the threefold position of the Fe (110) substrate. 3.3.

Free Energy of Activation

Activation energy is an important parameter used to estimate and predict a reaction. In this study, the thermodynamic integration 61 method has been used to examine the activation energy of the dissociation processes observed in the standard molecular dynamic simulations. The method allows us to calculate efficiently the reaction barrier when the collective variable is only needed to sample along the reaction coordinate.

Figure 14. Optimized structures and reaction coordinates for the dissociation mechanism of NBO atom at (a) 1100 K and (b) 1400 K and the depolymerisation (c) at 1500 K of the silicate cluster. Sodium ions have been removed for better illustration. Figure 14 shows the selected reaction coordinate ξ for the dissociation of NBO at 1100 K, 1400 K, and the depolymerisation process at 1500 K. The reaction coordinate for the reaction at 1100 K is more complicated than those at 1400 K and 1500 K. Considering the formation of the additional BO/Ob2 and to make sure the Si-Ob will not be broken during the constrained AIMD simulation, we add two additional collective variables r2 and r3 (Figure 14a). In order to determine the free-energy ACS Paragon Plus Environment

21


Page 22 of 32

profile, free-energy gradients (constrained force) for several samples are calculated. Starting from the selected configuration in the previous simulations (Figure 6a, 6f, and 10b), we performed a full relaxation to obtain the first minimum. The constrained bonds were then stretched from their equilibrium bond lengths. The free-energy gradient was evaluated from 2500 AIMD steps after the equilibration was established (at least 2500 AIMD steps) at 300 K. In order to obtain more accurate energy the time step was reduce to 0.25 fs. The barrier energy is obtained by calculating the path integral over the constrained force along the reaction coordinate connecting the initial and transition state (Equation 1) 61. The saddle point of free-energy can be defined by vanishing the constrained force, trapezoidal rule has been adapted to perform numerical integrations.

Δ = ξ1 ξ ξ ξ2

ξ∗

(1)

∗ is the free-energy gradient at a fixed value of reaction coordinate ξ∗ . ξ ξ

Figure 15. Free energy profile ∆E along the reaction coordinate (ξ) for the Si-NBO bond dissociation at 1100 K (square), 1400 K (dot), and the depolymerisation at 1500 K (triangle). The optimized structures show that the Si-O bond length in the configuration at 1400 K and 1500 K are 1.71 Å and 1.75 Å, respectively. However, the Si-O bond in the configuration at 1100 K is much longer (1.83 Å) indicating a stronger effect of the surface to the bond strength. Figure 15 shows the calculated free energy profile along the reaction paths the free energy gradient is provided in the supporting information (Figure S11). For the dissociation at 1100 K, the activation barrier is 0.52 eV. Meanwhile for the dissociation at 1400 K, the energy barrier is increased strongly to 0.81 eV. The reaction for the depolymerisation has the highest barrier energy of 0.89 eV. ACS Paragon Plus Environment

22

Page 23 of 32 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


The change of the barrier energy also reflects the impact of iron surface on the dissociation processes as mentioned in the previous sections. At 1500 K, the depolymerisation is not affected by the surface, hence the barrier energy is very high (0.89 eV) which implies a low reaction rate. However, due to the direct bond between NBO and Fe, the dissociation of the Si-NBO bond at 1400 K is easier with the barrier energy of 0.81 eV. At 1100 K, the effect of the iron surface becomes clearer with a strong elongation of the Si-NBO bond length. In addition, the formation of the additional BO bond further reduces the barrier energy to 0.52 eV, which is 42% lower than that of the depolymerisation. In general, the results of activation energy are in good agreement with the previous standard AIMD simulations, which indicates the Si-NBO bond dissociation is more favorable than the depolymerisation. In addition, the results also confirm the significant contribution of the additional bridging bond on the Si-NBO bond dissociation at low temperature.

4. DICUSSIONS The BOP values of Fe-O bonds in Table 2 not only prove the covalent characteristic of the Fe-O bond but also demonstrate that these bonds are very strong. Their strength is even stronger than Si-O bonds, which confirms the adhesion capability of the sodium silicate to metal surface. It is well-known that adhesion is one of the most important properties of high temperature lubricants since their antiwear performances consistently depend on the adhesion capability 11. The better the adhesion capability of a lubricant gets, the greater its antiwear performance achieves. The adsorption of sodium silicate at 1000 K indicates the number of NBOs involved in the bonding increases gradually during the process, which in turn improves the adhesion property of silicate lubricant. Furthermore, our simulation shows that the Na6Si2O7 prefers to adhere on the Fe surface through its NBOs, whereas BO does not form a stable bond with the iron surface. This might be because the BO is less chemically active than NBO 62. The result suggested that increasing the concentration of NBO may facilitate the reaction of Na6Si2O7 with Fe surface and eventually improve the adhesion of the lubricant. From the standard view, silicate glass has been constructed from the tetrahedral SiO4 units connected through Si–O–Si bridging bonds that are strong and directional. There are two types of cation in the silicate matrix. The first one is the basis of tetrahedron known as network former (Si4-), the network former cations help building the glassy network and increase the degree of polymerisation. The second one called network modifier (Na+) has an opposite effect, they can break up Si-BO links and play a role as the charge compensation. By adding more network modifiers such as alkali metals, the structure of silicate glasses can be degraded creating sheets, rings and short-chains such as sorosilicate or orthosilicate molecules. Therefore, the concentration of NBO can be adjusted by ACS Paragon Plus Environment

23


Page 24 of 32

modifying the concentration of sodium (network modifier) in the lubricant compound 63. However, the depolymerisation process may lead to an unstable network that has a negative effect on the antiwear capability. It is well known in phosphate lubricants that the increasing degree of polymerisation will result in a much tighter network, which is believed to be responsible for an improvement of chemical and mechanical stability of phosphate tribofilm, and reducing wear

24, 64

. Furthermore, increasing

network modifier concentration can break the glassy network and reduce the degree of polymerisation, which in turn lower the viscosity

65

. Since the reaction rate, tribofilm stability and viscosity can be

controlled by the concentration of network modifier, finding the optimal ratio of network modifier or NBO is worth further investigating. As mentioned in the simulation at 1100 K, the formation of Si1-O5/Ob2 bond transforms the NBO (O5) to the second BO (Ob2). The new silicate structure called 2-Ring defect, which is a set of edgesharing SiO4 tetrahedra. The 2-Ring defective structure has been identified experimentally during the fracturing of silica

66-68

. In our simulation, the formation of 2-Ring defective structure eliminated one

NBO that later becomes adsorbed on the Fe surface. Therefore, this process can weaken the adhesion capability by reducing the number of Si-O-Fe linkages. In addition, the dissociation of NBO will create free oxygens in the tribofilm that leads to the reduction of anti-oxidation ability of the lubricant. The oxidation process not only causes a huge loss of material but also has indirect effect into the anti-wear performance of the lubricant. This is because during the shearing process, the scale oxide layer on the top can be abraded creating iron oxide particles, so oxygen can attack the nascent iron surface 69. The iron oxide particles are relatively hard with high melting point (>1200 °C) and can cause severe damage to the film. As the shearing process continues, more nascent surface areas are exposed to oxygens, as a consequence the wear will increase over the time. Therefore, in order to reduce loss and wear caused by oxidation, we have to eliminate or reduce the dissociation of NBO. One possible solution is to dope high coordination number cations as network formers such as Al(6) or Fe(5) into the glass. Hence, the new network formers can have additional bonds with NBOs when 2-Ring defects are formed without eliminating the other NBOs. It is also believed that the incorporation of high coordination number network former like Fe can create cross-linking that consequently increase the degree of network and reduce wear. Both PDOS and DDEC charges indicate a large amount of charge is transferred from O to the Fe layers. The phenomena may contribute to the formation of the tribofilm in several aspects. At first, the charge on the Fe surface becomes more and more negative as the number of bonds with NBOs increases. As a result, the negatively charged surface may create an electrostatic field and attract positively charged atoms thanks to the Coulombic interactions. It is well-known that the mobility of ACS Paragon Plus Environment

24

Page 25 of 32 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


alkali ion in aqueous silicate is extremely high due to the weak ionic bond with NBOs 70-71. In addition, as the charge transfer from oxygen to the surface also reduces the Coulombic interaction between O and Na, the mobility of Na may become even higher after the silicate adsorption. Therefore, there is a possibility that sodium atoms will respond to the electrostatic field and move toward the Fe surface. In fact, high concentration of alkali elements on steel surface has been found in several glassy lubricants. For instance, Tran et al.

72

reported a bi-layer tribofilm consisting of B2O3 on top of a rich Na layer

when doing the test with sodium borate lubricant. A high concentration of Na on the top of iron oxide surface has been also found in phosphate lubricant 64. Furthermore, our simulation also suggested that the dissociation of NBOs may lead to the formation of the direct Si-Fe bonds. The process can bring an opposite effect to the adsorption of NBOs since Si atoms will drain electrons from the surface. The flow of alkali ions may results in a layering processing in the tribofilm. In particular, a high level of polymerisation can be observed in the layer where alkali ions are drained. Meanwhile, the rich alkali ion layer, which is likely located near the surface, will have high degree of depolymerisation. Unfortunately, at the moment there is no experimental publication that clearly identifies the composition of sodium silicate tribofilm. However, layering process has been observed in phosphate and borate lubricants. For example, Wan et al.

64, 73

found a tribofilm with different chain lengths of

polyphosphate and has a gradient change in composition. According to Wan’s report, the top layer is short chain-length polyphosphate which is mainly caused by oxide the long chain-length polyphosphate when the tribofilm is exposed to the air; the sublayer is composed of relatively long chain-length polyphosphate on top of short chain-length polyphosphate. The result, therefore, indicate a depolymerisation process in the bottom layer while the upper layer show a higher degree of polymerisation. In addition, a polymerisation process of B2O3 on top of rich Na layer has been observed experimentally 72. The simulation at 1500 K indicates that extremely high temperature can cause serious damage to the lubricant. The lubricant is decomposed into small fragments that may results in the instability of the tribofilm. As a consequence, high wear and fiction are expected to obtain. Matsumoto et al.

2

reported an increase in fiction coefficient of sodium silicate lubricant which is likely caused by the decomposition of the lubricant at high temperature. The present study does not account for the effect of pressure and shear which are believed to play an important role in the formation of tribofilm such as the polymerisation by pressure

24

. Furthermore, a layer of Fe2O3 is normally built as a consequence of

oxidation of nascent iron with oxygen in the environment or free oxygen in lubricants. Therefore, it is necessary to investigate the reaction between silicate lubricants with the iron oxide surface. The effect


25


Page 26 of 32

of pressure, shear as well as reaction mechanism with iron oxide surface will be reported in the future work.

3. CONCLUSIONS FPMD simulations and DFT calculations have been conducted to study the adsorption and dissociation of sodium sorosilicate (Na6Si2O7) on Fe (110) surface. The comparisons of adsorption processes and electronic structure of some selected configurations at different temperatures have provided detailed insights into the bond nature of the tribology system, the effect of the iron surface as well as temperature on the adhesion of silicate lubricant. The findings in the present study can be summarized as follows: (i)

Silicate lubricant adheres on the Fe (110) surface through multiple Si-O-Fe linkages. Iron surface has significant effect on the strength of Si-O bonds in the Si-O-Fe linkages. In particular, Fe can drain electrons from NBOs leading to the decrease of Si-NBO bond strength. Therefore, NBOs are much easier to be dissociated to create free oxygens in the tribofilm. That in turn may reduce the anti-oxidation ability of the tribofilm. In addition, the charge transfer from silicate lubricant to the iron surface will modify the net charge of the surface to negative. The result suggested that an electrostatic filed can be created on the surface and attracts alkali ions.

(ii)

Temperature plays a crucial role in the dissociation of the lubricant. At low temperature (

First-Principles Study of the Adsorption and ... - ACS Publications

Recommend Documents